Have audit-users.py write out serialized data

This allows us to "query" the datastructure for different perspectives
without needing to rerun the costly queries each time we update
audit-uses.py. The script is predominantly collecting data now, then we
can use the python repl or other scripts to give us better insights.

We also do a small refactoring to simplify the collection of data.

Change-Id: Ie777ae706050b38ce294a1acf9b1b843fcf5ab41

tools/gerrit-account-inconsistencies/audit-users.py

@@ -35,6 +35,7 @@ import datetime
 import json
 import getpass
 import requests
+import yaml
 
 TIME_FORMAT = '%Y-%m-%d %H:%M:%S.%f'
 TODAY = datetime.datetime.now()
@@ -127,6 +128,34 @@ def check_recent_changes(account_id, account_info, auth):
     else:
         account_info['recent_review'] = None
 
+def gather_user_info(account_id, user, auth):
+    detail = get_account_detail(account_id, auth)
+    if 'registered_on' in detail:
+        user[account_id]['registered_on'] = detail['registered_on']
+    if 'username' in detail:
+        user[account_id]['username'] = detail['username']
+    sshkeys = get_account_sshkeys(account_id, auth)
+    if sshkeys:
+        user[account_id]['sshkeys'] = True
+    eids = get_account_externalids(account_id, auth)
+    for eid in eids:
+        # We only care about login.ubuntu urls now
+        if 'login.ubuntu' in eid['identity']:
+            r = requests.head(eid['identity'])
+            if r.status_code == 200:
+                # If there is an openid and it is valid we add it
+                # to the list of valid openids
+                user[account_id]['openids'].append(eid['identity'])
+            else:
+                user[account_id]['invalid_openids'].append(eid['identity'])
+
+    check_recent_changes(account_id, user[account_id], auth)
+
+    if user[account_id]['recently_used']:
+        user['recently_used'].append(account_id)
+    else:
+        user['nonrecently_used'].append(account_id)
+
 def get_user_activity(users, auth=None):
     for email in users.keys():
         users[email]['active'] = []
@@ -148,62 +177,21 @@ def get_user_activity(users, auth=None):
                                         'active': True,
                                         'username': None,
                                         'sshkeys': None,
-                                        'openids': []}
+                                        'openids': [],
+                                        'invalid_openids': []}
             users[email]['active'].append(account_id)
+            gather_user_info(account_id, users[email], auth)
 
-            detail = get_account_detail(account_id, auth)
-            if 'username' in detail:
-                users[email][account_id]['username'] = detail['username']
-            sshkeys = get_account_sshkeys(account_id, auth)
-            if sshkeys:
-                users[email][account_id]['sshkeys'] = True
-            eids = get_account_externalids(account_id, auth)
-            for eid in eids:
-                # We only care about login.ubuntu urls now
-                if 'login.ubuntu' in eid['identity']:
-                    r = requests.head(eid['identity'])
-                    if r.status_code == 200:
-                        # If there is an openid and it is valid we add it
-                        # to the list of valid openids
-                        users[email][account_id]['openids'].append(eid['identity'])
-
-            check_recent_changes(account_id, users[email][account_id], auth)
-
-            if users[email][account_id]['recently_used']:
-                users[email]['recently_used'].append(account_id)
-            else:
-                users[email]['nonrecently_used'].append(account_id)
         for account in inactive_j:
             account_id = str(account['_account_id'])
             users[email][account_id] = {'recently_used': False,
                                         'active': False,
                                         'username': None,
                                         'sshkeys': None,
-                                        'openids': []}
+                                        'openids': [],
+                                        'invalid_openids': []}
             users[email]['inactive'].append(account_id)
-
-            detail = get_account_detail(account_id, auth)
-            if 'username' in detail:
-                users[email][account_id]['username'] = detail['username']
-            sshkeys = get_account_sshkeys(account_id, auth)
-            if sshkeys:
-                users[email][account_id]['sshkeys'] = True
-            eids = get_account_externalids(account_id, auth)
-            for eid in eids:
-                # We only care about login.ubuntu urls now
-                if 'login.ubuntu' in eid['identity']:
-                    r = requests.head(eid['identity'])
-                    if r.status_code == 200:
-                        # If there is an openid and it is valid we add it
-                        # to the list of valid openids
-                        users[email][account_id]['openids'].append(eid['identity'])
-
-            check_recent_changes(account_id, users[email][account_id], auth)
-
-            if users[email][account_id]['recently_used']:
-                users[email]['recently_used'].append(account_id)
-            else:
-                users[email]['nonrecently_used'].append(account_id)
+            gather_user_info(account_id, users[email], auth)
 
 
 if __name__ == '__main__':
@@ -216,6 +204,9 @@ if __name__ == '__main__':
 
     users = read_email_list()
     get_user_activity(users, auth=auth)
+    with open('audit-results.yaml', 'w') as f:
+        yaml.dump(users,  default_flow_style=False, explicit_start=True,
+                  indent=4, stream=f)
 
     # TODO there are probably better ways to present this data.
     print()